PRIQ questions for VOIP in multiwan/multivlan



  • Hello,

    I'm new with traffic shapping, doing a small test lab before going in a production case.

    I have VOIP, phones and IPBX are on the same vlan. IPBX is talking to the provider with a siptrunk.
    Almost all other VLANs are dedicated to customers (1VLAN/customer) it's a small business center.

    My need: give priority to VOIP from my IPBX to the provider

    What I did: After many readings on the forum and also the online book (I subscribed to the book for this point) I decided to follow the Wizard "traffic_shaper_wizard_multi_all" and go for PRIQ.

    My questions:
    1. I have 2 VDSL, bandwidth is not fix and both VDSL have different values. As the bandwidth values are not stable what do I have to declare in the wizard "Shaper configuration"?

    • Maximum values
    • Average values
    • Minimum Values

    2. During my tests I noticed that VOIP trafic is going in the good Queue even if I did not mentionned the upstream server in the wizard. The floating (there is only one when no server is declared in the wizard) does not refer to any port, protocol.
    After I added the IP of my server in the wizard a second floating rule appeared and the IP appeared in source in a floating rule and in destination in the other floating rule.
    My question: Based on what does PFSense determine what is VOIP traffic?

    I tried to do short, if you need more informations to answer do not hesitate.

    Thanks for your informations/ideas/experience share



  • pfSense can't know what you current effective bandwidth is. You need to tell it a fixed value. If your ISP can't provide that, there's not much you can do. You'll hyst to pick a value that works best for you, by your own definition of best. If you can't get stable bandwidth, then you'll have to trade stability for bandwidth. Either set the bandwidth so low you don't get stability issues or leave your bandwidth high and deal with the latency.

    pfSense has a bunch of default rules it creates. Look under the floating rules if you want to see what the definitions are.



  • Hello Harvy,

    First, thanks for taking time to answer.

    Question1:

    pfSense can't know what you current effective bandwidth is. You need to tell it a fixed value. If your ISP can't provide that, there's not much you can do.

    OK, understood

    You'll hyst to pick a value that works best for you, by your own definition of best. If you can't get stable bandwidth, then you'll have to trade stability for bandwidth. Either set the bandwidth so low you don't get stability issues or leave your bandwidth high and deal with the latency.

    As PRIQ is working with priority, do I really have to tell PFSense what "amount" of bandwidth I have?
    In the documentation, about PRIQ and VOIP (in the wizard section) it is said

    Leave Upload and Download blank if using PRIQ,

    In a PRIQ context, what is the use of giving upload/download values to the connection? (is there a triger?)

    What wrong will happen if I put values too high or too low?

    Question 2:

    pfSense has a bunch of default rules it creates. Look under the floating rules if you want to see what the definitions are.

    : yes that's what I did and that's the point of my question: If I do not put the IP of the server of the provider, there is only one floating rule created. And that rule do not describe port or protocols to match. Despite this, traffic is using the good queue (with no port, protocol, IP on the Internet). How do you explain this magic?

    Thanks



  • Think of it like; PRIQ itself doesn't need bandwidth values but ALTQ does, which PRIQ is built within. The traffic-shaping interface (WAN, LAN, etc) requires the bandwidth to be set for successful traffic-shaping.

    Practically, a universally fundamental aspect of QoS/traffic-shaping is that whichever device is doing the traffic-shaping must be the bandwidth bottleneck, which is accomplished by configuring artificially lowered bandwidth values on the traffic-shaping interface.



  • If this is right and if I corectly understood:
    https://calomel.org/pf_hfsc.html

    CBQ and PRIQ type queues with a higher priority are served first if the link is saturated and the "realtime" bandwidth is also exhausted.

    • PRIQ is doing the job from the moment the queues are over the size of the connections
    • Queues systems, when in use induce latencies
    Don't forget I'm on VDSL, asymetric connection without garenteed (?) bandwidth
    

    Let's get the following exemple:
    Good days: 75Mbps Down/6MbpsUp.
    Average:  50Mbps Down/5MbpsUp.
    Bad days:  25Mbps Down/2,5MbpsUp.

    • If the connection speed is set to high value 75Mbps Down/6MbpsUp and the real value of the connection is under (for example 50Mbps Down/5MbpsUp):
      – If my needs for connection are under the real connection capacity (50Mbps Down/5MbpsUp. ) : no problem
      -- If my needs for connection are between the real connection capacity and under the value of my connection as declared (75Mbps Down/6MbpsUp): problem: PRIQ will not trigger and my VOIP can not function with quality. The network is consuming all the bandwidth so VOIP will sufer, the priq will not work because it will only work when my network is filling connection queues with 75Mbps Down/6MbpsUp
      -- If my needs for connection are over my connection as declared (75Mbps Down/6MbpsUp) : PRIQ will triger, VOIP will work with may be a bit latency due tu queues system.

    So IF I set my connection too high I risk PRIQ not to work when the needs for connection are between the triger and the realtime values of the connection.
    If I set my connection too low, trigger will often work (when not needed) and PRIQ will work (when not needed) and induce latencies.

    Is it right?


Log in to reply